Limited availability of sawdust and planer shavings and an increasing demand for biomass pellets in Europe are pushing the market toward other, more problematic raw materials with broader variation in total fuel ash content and composition of the ash forming elements as well as in their slagging tendencies. The main objective in the present work was therefore to determine the influence of fuel−ash composition on residual ash and slag behavior. Twelve different biomass pellets were used: reed canary grass (two different samples), hemp (two different samples), wheat straw, salix, logging residues (two different samples), stem wood (sawdust) as well as spruce, pine, and birch bark. The different pellet qualities were combusted in a commercial under fed pellet burner (20 kW) installed in a reference boiler. Continuous measurements of O2, CO, CO2, HCl, SO2, and total particle matter mass concentrations were determined in the exhaust gas directly after the boiler. The collected slag deposits, the corresponding deposited bottom ash in the boiler and the collected particle matter were characterized with X-ray diffraction (XRD) and scanning electron microscopy combined with energy dispersive X-ray analysis (SEM/EDS). For biomass fuel pellets rich in silicon (either inherent or contaminated with sand) and low content of alkaline earth metals the main part of the potassium reacted with the silicon rich ash-residual, forming sticky alkali−silicate particles, which were not entrained from the burner and thereby giving rise to/initiating slag formation. Silicon rich fuels, i.e. fuels were the ash characteristics were dominated by silicate−alkali chemistry, therefore generally showed relatively high slagging tendencies. Straw fuels have typically this ash composition but exceptions to these general trends exists (e.g., one of the hemp fuels used in this work). Wood derived fuels with a relatively low inherent silicon content therefore showed low or relatively moderate slagging tendencies. However, contamination of sand material to these fuels may greatly enhance the slagging tendencies.

In this paper, a comparison between four different types (both empirical and theoretical) of techniques to predict the slagging tendencies in residential pellet combustion appliances was performed. The four techniques used were the standard ash fusion test (SS ISO-540) used in the Swedish pellet standard (SS 18 71 20), thermal analysis (TGA/DTA), thermochemical model calculations, and a laboratory-scale sintering test. The tests were performed with 12 pelletized biomass raw materials, and the results were compared with measured slagging tendencies in controlled combustion experiments in a commercial under-fed pellet burner (20 kW) installed in a reference boiler. The results showed significant differences in the prediction of slagging tendencies between different predicting techniques and fuels. The method based on thermal analysis (TGA/DTA) of produced slags must be further developed before useful information could be provided of the slagging behavior of different fuels. The used sintering method must also be further improved before the slagging tendency of fuels forming slags extremely rich in silicon (e.g., some grasses) can be predicted. Relatively good agreement was obtained between results from chemical equilibrium calculations and the actual slagging tendencies from the combustion tests. However, the model calculations must be further improved before quantitative results can be used. The results from the standard ash fusion test (SS ISO 540) showed, in general, relatively high deformation temperatures, therefore predicting a less problematic behavior of the fuels in comparison to the actual slagging tendencies obtained from controlled combustion experiments in commercial pellet burner equipment. Nevertheless, the method predicted, in most cases, the same fuel-specific slagging (qualitatively) trends as the corresponding combustion behavior.